A new classification tool helped guide the treatment of bacteremic patients while clinicians awaited antibiotic resistance results, investigators reported.
The clinical decision tree had a positive predictive value of 91% and a negative predictive value of 92% for determining whether certain gram-negative infections produced extended-spectrum beta-lactamase (ESBL), Catherine Goodman, PhD, of the Johns Hopkins Bloomberg School of Public Health, Baltimore, and her associates wrote online in Clinical Infectious Diseases. “These predictions may assist empiric treatment decisions in order to optimize clinical outcomes while reducing administration of overly broad antibiotic agents that can select for further resistance emergence,” they added.
CDCKlebsiella pneumoniae
Bacteria that produce ESBL can hydrolyze all broad-spectrum beta-lactam antibiotics except carbapenems. Rapid tests for beta-lactamase genes can shorten the lag time between gram-stain identification and antimicrobial resistance results, but are cost prohibitive for most clinical laboratories and often do not assess ESBL gene groups, the researchers said. To find a way to predict which infections are characterized by ESBL production, they studied adults hospitalized at Johns Hopkins from October 2008 to March 2015 with bloodstream isolates of Klebsiella pneumoniae (40% of patients), Klebsiella oxytoca (4% of patients), and Escherichia coli (56% of patients). Most bacteremias began as urinary tract infections (34% of cases), followed by intra-abdominal infections (24%), catheter-related infections (16%), and biliary infections (14%) (Clin Infect Dis. 2016 Jul 26. doi:10.1093/cid/ciw425).
A total of 194 patients (15%) had bacteremias that produced ESBL, according to the investigators. Using a technique called binary recursive partitioning, they compared these patients with ESBL-negative patients to create a clinical decision tree based on five yes-or-no questions. The tree first asked if the patient had been colonized or infected with ESBL-producing bacteria within 6 months, and if so, whether the patient currently had an indwelling catheter. Patients meeting both criteria had a 92% chance of being ESBL positive. Patients with a recent history of ESBL but no catheter had an 81% chance of being ESBL positive if they were at least 43 years old, but a 75% chance of being ESBL negative if they were under age 43 years.
Among patients with no recent history of ESBL, the decision tree asked about hospitalization in a country with a high ESBL burden and antibiotic therapy during the past 6 months. Patients responding “yes” to both questions had a 100% chance of being ESBL positive. Patients with only the geographic risk factor had a 63% chance of being ESBL negative, and patients with neither risk factor had a 93% chance of being ESBL negative.
The decision tree detected only half of ESBL cases because there was a subgroup with no recent ESBL history or geographic exposure, the investigators noted. “The poor predictive nature of health care–associated variables within this patient subset may suggest a high proportion of community-acquired ESBL infections. Indeed, although risk factors for ESBLs have traditionally focused on the health care setting, increasing reports describe the community as an important ESBL reservoir,” they added. Nonetheless, of 194 patients with ESBL bacteremia, 35% received empiric carbapenem treatment within 6 hours after identification of the bacterial genus and species, the investigators emphasized. “Utilization of the decision tree would have increased ESBL case detection during the empiric treatment window by approximately 50%.”
The National Institutes of Health funded the study. The researchers reported having no conflicts of interest.